1. SEMANTIC AGENT SYSTEMS Towards a Reference Architecture for Semantic Agent Systems Applied to Symposium Planning Usman Ali UNB FCS,Fredericton, NB 1

2. Outline 2  Background  Agent Scenario  Virtual Organization  Organizational Designs  Semantic Agent Systems  Evaluation/Comparison  Conclusion

3. Semantic Web Vision 3

4. Agent Scenario Consider a Web- enabled method for saving the doomed crew of The Perfect Storm. How could Web agents have helped? 4 James Hendler: Agents and the Semantic Web. IEEE Intelligent Systems Journal (March/April 2001). http://www.cs.rpi.edu/~hendler/AgentWeb.ht mlhttp://www.cs.rpi.edu/~hendler/AgentWeb.ht ml

5. Virtual Organization "Virtual Organizations are a set of individuals and institutions that need to co-ordinate resources and services across institutional boundaries". 5 N. Criado, E. Argente, V. Juli´an, V. Botti: Designing Virtual Organizations. 440-449, In: Demazeau, Y.; Pavón, J.; Corchado, J.M.; Bajo, J. (Eds.): 7th International Conference on Practical Applications of Agents and Multi-Agent Systems (PAAMS'09), Series: Advances in Intelligent and Soft Computing, Vol. 55, 2009.

6. Software Personal Assistants Software personal assistants (SPA) are an active research area that one day might change the face of our human organizations. Organizational Structures Star Ring Mixed/Random 6 Towards Agent-Oriented Conceptualization and Implementation / Pratik K. Biswas http://www.igi-global.com/bookstore/titledetails.aspx?titleid=76

7. Agent Centered Versus Organization Centered Approach Agent Centered States of an agent and of the relation between these states and its overall behaviour. Organization Centered Concepts of ‘organizations’, ‘groups’, ‘communities’, ‘roles’, ‘functions’, etc. play an important role. 7 Jacques Ferber, Olivier Gutknecht, and Fabien Michel: From Agents to Organizations: an Organizational View of Multi-Agent Systems. Agent-Oriented Software Engineering (AOSE) IV, P. Giorgini, Jörg Müller, James Odell, eds, Melbourne, July 2003, LNCS 2935, pp. 214-230, 2004

8. Organization Design Open Agent Architecture (OAA):  Flexible, dynamic communities of distributed software agents.  Human users and software agents, in an OAA, express their requests in terms of, “What will be done?” rather than “How will it be done?” 8 In a distributed software architecture, sharing information or interaction has to be predefined which makes it a rigid distributed architecture environment. http://www.ai.sri.com/oaa/http://www.ai.sri.com/oaa/ {Open Agent Architecture: Technical White Paper}

9. Semantic Agent Systems 9 Semantic Agent Systems use rules and Ontologies for creating virtual organizations as multi-agent systems that can support collaborative teams on the Semantic Web. They provide the infrastructure for rule- and ontology-based collaboration between the distributed members of such a virtual organization.

10. Multi Agent Frameworks Academic World Business World 10

11. EMERALD 11 Kalliopi Kravari, Taylor Osmun, Harold Boley and Nick Bassiliades, Cross-Community Interoperation Between the EMERALD and Rule Responder Multi-Agent Systems.

12. RULE RESPONDER 12 http://ruleml.org/RuleResponder/

13. Organizational Agent  The organizational agent represents the goals and strategies shared by each committee chair.  It contains rule sets that describe the policies and regulations of the RuleML Symposium.  Delegates incoming queries to the chair’s PAs. 13

14. Personal Agent A personal agent assists a single chair of the symposium,(semi-autonomously) acting on his/her behalf. Each personal agent contains a rule- base FOAF-like profile. It contains a FOAF*-like fact profile plus FOAF-extending rules to encode selected knowledge of its human owner. 14

15. External Agent External agents exchange messages with the OA. They submit queries and receive answers. End users, as external agents, interact with the OA using a Web (HTTP) interface to the Symposium Planner. Support for simultaneous external agents. Many EAs can communicate with the OA. 15

16. Rule Engines Prova is mainly used to realize the organizational agents of Rule Responder It implements Reaction RuleML for agent interaction (event-condition-action rules) 16 OO jDREW is used to realize the personal agents of Rule Responder It implements Hornlog RuleML for agent reasoning (Horn logic rules) Supports rules in two formats: POSL: Positional Slotted presentation syntax RuleML: XML interchange syntax

17. COMMUNICATION MIDDLEWARE Mule Enterprise Service Bus (ESB)  Mule is used to create communication end points at each personal and organizational agent of Rule Responder.  Mule supports various transport protocols (i.e. http, JMS, soap)  Rule Responder uses http and JMS as transport protocols. 17

18. MULE ENTERPRISE SERVICE BUS 18

19. Reaction RuleML  Reaction RuleML is a branch of the RuleML family that supports actions and events.  When two agents want to communicate, each others’ Reaction RuleML messages are sent through the ESB.  The ESB carries RuleML queries (requests), answers (results), and rule bases to/from agents. 19

20. 20

21. Online Use Case Demo http://www.defeasible.org/ruleml2011ijcai/?q=node/25 http://de.dbpedia.org/redirects/ruleml/ACE2ReactionRuleML/index.jsp 21

22. 22

23. 23

24. Evaluation of Semantic Agent Systems 24 RREM Separate rule and ontology languages, or combine them in a hybrid or homogeneous manner Distinguish a performative (pragmatic) layer from a content (semantic) layer Organise agent communication in a hierarchical (client-server) or networked (peer-2-peer) or reconfigurable (dynamic) fashion. Can be built on a multi-agent platform such as JADE or be based on an Enterprise Service Bus such as Mule. Can be used to define autonomous agents or semi- autonomous agents. Can operate on a private intranet, a cross-organisational extranet or the open Internet. Can realise virtual organisations in which agents solve various kinds of tasks, including document retrieval, question answering, and information integration. Rule Responder uses both Ontologies and Rules. Rule Responder uses FIPA and Reaction RuleML. Rule Responder uses hierarchical architecture. Rule Responder uses Mule ESB. Emerald uses JADE. Rule Responder and Emerald provide semi autonomous agents. Rule Responder can operate locally as all resources are local. Emerald has to access third party resources. Rule Responder provides agents (OA, PA, EA, CA) to perform different tasks. Emerald provides Knowledge Customizable (KC)-Agents.

25. Conclusion 25